29
kE
= 1010-2l./(mol sec), as r e p ~ r t e d . ~ ~To ’ ~ bring the
values into agreement unilaterally would require that either be in error by 1.4 kcal or that kE be only 108.7 l./(mol sec). This latter is improbably low but may be taken as a lower limit for the recombination of ethyl radicals. It suggests that the main source of disagreement is cumulative errors in the thermodynamic parameters. is about f 1.6 kcal, which The uncertainty in is well within the above limits and shows very good 10% of the propane yields from which K M E (kinetic) was determined. Furthermore, any such adjustments of this sort should also attempt to take into account propane and butane consumed during the course of reaction. That the effects of such side reactions are small and perhaps even self-compensating is evident from the invariance of KME(kinetic) with extent of reactions (Table I). (19) Taking the average of the values at 50 and 100” in ref 6 and assuming that the recombination has no activation energy.
agreement between calculated and observed values. The chief contribution to these uncertainties in AGO are the heats of formation of the two radicals. The AHof and So values for Me1 and Et1 are much better known and have between them a combined uncertainty which amounts to less than 0.5 kcal contribution to AGO. Fessenden’s reported values7 for k E obtained by direct esp measurement of Et. disappearance, albeit in liquid ethane, extrapolate to l./(mol sec) at 415°K. This yields a KME(kinetic) of 12, in substantially better agreement with the thermodynamic calculation. Both Shepp and Kutschke6 and Fessenden7 report activation energies for Et. E t - of 2.0 and 0.8 kcal/ mol, respectively. While the present data have no bearing on this aspect, the desirability of further measurements at lower temperatures is indicated.
+
Steric Effects on Phosphorus Hyperfine Coupling Constants in Nitroaromatic Anion Radicals’ Wilson M. Gulick, Jr.
Contribution from the Departments of Chemistry, Cornell University, Ithaca, New York 14850, and the Florida State University, Tallahassee, Florida 32306. Received February 27, 1971 Abstract: Electron spin resonance spectra of the anion radicals produced by electrolyticreduction of 4-nitrophenyl diphenylphosphinate (I), 2,6-dimethyl-4-nitrophenyldiphenylphosphinate (11), 2,6-diphenyl-4-nitrophenyldiphenylphosphinate (111), 0,O-diphenyl S-Cnitrophenyl thiophosphate (IV), and diethyl 2,6-dimethyl-4-nitrobenzylphosphonate (V) have been observed. The magnitude of the phosphorus hyperfine coupling constant, UP, varied from 6.07 G for the anion radical obtained from I in acetonitrile to 27.15 G for the anion radical of IV in dimethylformamide. In the series of similar radicals 1-111, U P increases with increased bulk of the substituents in the 2 and 6 positions on the nitroaromatic ring. These data, in addition to comparison of data for V with data reported previously for its unmethylated analog, support the suggestion made previously that the magnitude of UP contains an angular dependence similar to that found for proton hyperfine couplings when spin transmission occurs via hyperconjugation.
tereochemical effects upon * ‘P hyperfine coupling constants in nitroaromatic anion radicals have been interest to us for some time. In a previous report2 we described the electron spin resonance (esr) spectra of seven such radicals obtained by one electron reduction of phosphates and a phosphonate. We report now the extension of that work t o a new series of nitroaromatic phosphorus anions derived from compounds I-v . Our previous results2 suggested that steric perturbations of the bonds linking the phosphorus atom to the nitroaromatic moiety (P-X-C) are important in determining the magnitude of the hyperfine interaction. The results reported now substantiate and elaborate this suggestion and provide sufficient data to permit some semiquantitative estimates of the parameters which describe ap.
S of
(1) Presented in part at the 158th National Meeting of the American Chemical Society, New York, N. Y., Sept 1969. Address correspondence to the Department of Chemistry, Florida State University, Talla-
hassee. Fla
(2j-W.-M. Gulick, Jr., and D. H. Geske, J . Amer. Chem. SOC.,88, 2928 (1966).
Experimental Section Solvents for electrochemistry and tetraethylammonium perchlorate supporting electrolyte were commercially available and purified accordjng to established literature procedures. Diphenyl chlorophosphate, diphenylphosphinic acid, triethyl phosphite, and 4-nitrothiophenol were obtained from Aldrich Chemical Co. Mucobromic acid, 1,3-diphenyl-2-propanone,triethylamine, and thionyl chloride were supplied by Eastman Organic Chemicals and 3pentanone came from Matheson Coleman and Bell. 2,6-Dimethylbenzoic acid was obtained from K & K Laboratories.
Gulick
1 Phosphorus HyperJne
Coupling Constants
30 4Nitrophenyl Diphenylphosphinate (I). A sample of this material was the gift of Professor Paul Haake and Dr. Gail Hursta and was used as received. 2,6-Dimethyl-4-nitrophenyl Diphenylphosphinate 01). Sodium nitromalonaldehyde monohydrate was prepared by reaction of sodium nitrite and mucobromic acid as described by Fanta.' 2,6Dimethyl-4nitrophenol was obtained by condensation of 3-pentanone with the nitromalonaldehyde according to the method of Hills as modified by Jones and Kenner.6 Diphenylphosphinic chloride resulted from reaction of diphenylphosphinic acid with excess thionyl chloride. The product was isolated and purified by vacuum distillation. Reaction of equimolar quantities (27 mmol) of diphenylphosphinic chloride and 2,6-dimethyl-4nitrophenol was carried out in acetone solution in the presence of approximately a 10 % molar excess of triethylamine. Triethylamine hydrochloride was removed by filtration and vacuum evaporation of the solvent afforded crude I1 in high yield. The crude material was twice recrystallized from 95 % ethanol. The resulting colorless to white crystals melted at 141-141.5". We find no previous mention of this compound in the literature. The composition is established by elemental analysis. And. Calcd for CzoH18N04P:C, 65.39; H, 4.94; N, 3.81; P, 8.43. Found:7 C, 65.64; H, 4.97; N, 3.64; P, 8.70. 2,6-Diphenyl-4-nitrophenyl Diphenylphosphinate (111). 2,6-Diphenyl-4-nitrophenol was prepared from sodium nitromalonaldehyde monohydrate and 1,3-diphenyl-2-propanoneas described by Hill5 Reaction of this phenol in acetone solution with an equimolar quantity of diphenylphosphinic chloride in the presence of triethylamine, as described for 11, gave approximately 90% yield of crude 111 as a yellow semisolid. This crude product was dissolved in ether. The solution was extracted with two portions of distilled water, dried with MgSOa, boiled briefly with decolorizing carbon, and filtered. Large, colorless crystals were obtained on cooling the ether solution. The product was recrystallized once more from ether. The resulting crystals contained ether which was removed by grinding the crystals to a fine powder and heating to 60" in uacuo. The resulting material melted at 158.5-159.5". Since 111 appears to be a new compound, the composition is confirmed by elemental analysis. Anal. Calcd for CB0HzzNO4P: C, 73.32; H, 4.48; N, 2.85; P, 6.31. Found:? C, 73.49; H , 4.62; N, 2.79; P, 6.42. 0,O-Diphenyl S-4Nitrophenyl Thiophosphate (IV). Reaction of diphenyl chlorophosphate and 4-nitrothiophenol in acetone solution in presence of excess triethylamine as described above afforded crude IV as a viscous oil upon removal of the solvent. This reaction did not proceed when ether was used as the reaction medium. The oil was dissolved in methanol. Slow evaporation of the methanol gave a pale brown solid which had an infrared spectrum consistent with the structure IV; in particular, the S-H stretching band was absent. Further recrystallization from alcohols failed; however pale yellow prisms were obtained by slow evaporation of a solution of IV in a mixture of 2 vol of hexane and 1 vol of chloroform. These crystals melted at 69.5-70.5'. We find no evidence of previous preparation of this compound. The composition is confirmed by elemental analysis. Anal. Calcd for C18H14OjNPS: C , 55.81; H, 3.64; N, 3.61; P, 8.00; S, 8.28. Found? C, 56.17; H , 3.65; N, 3.41; P, 7.98, 7.89; S, 8.50, 8.39. Diethyl 2,6-Dimethyl-4-nitrobenzylphosphonate (V). 2,6-Dimethylbenzoic acid was reduced with LiA1H4as described by Raaen and E a s t h a d to produce the benzyl alcohol. Treatment of the alcohol with thionyl chloride yielded 2,6-dimethylbenzyl chloride,O bp 55" (0.1 mm). Reaction of 10 g (64 mmol) of the benzyl chloride with excess triethyl phosphite according to the method of Kagan, et aI.,lO produced a nearly quantitative yield of diethyl 2,6dimethylbenzylphosphonate, bp 110" (0.1 mm). The composition was established cia infrared spectroscopy and high-resolution mass spectroscopy : molecular ion m/e 256.1225 (calcd for C13Hz103P, 256.1228). Nitration was carried out using a slight modification of (3) P.Haake and G. Hurst, J . Amer. Chem. SOC.,88, 2544 (1966). (4) P. Fanta, Org. Syn., 32,95 (1952). ( 5 ) H . B. Hill, Amer. Chem. J . , 24, l(1900). (6) E.Jones and J. Kenner, J . Chem. SOC.,1842 (1931). (7) Galbraith Laboratories, Knoxville, Tenn. (8) Schwarzkopf Microanalytical Laboratory, Woodside, N. Y. (9) V. F. Raaen and J. F. Eastham, J . Amer. Chem. SOC.,82, 1349
(1960). (10) F. Kagan, R. D. Birkenmeyer, and R. E. Strube, ibid., 81, 3026 (1959).
Journal of the American Chemical Society
94:l
the procedure used by KosolapofP for nitration of di-n-butyl benzylphosphonate. Diethyl 2,6-dimethylbenzylphosphonate (8.45 g, 33 mmol) was added dropwide to an icetold mixture of 8 ml of concentrated H2S04 and 4 ml of concentrated HNO1. After 40 min the reaction mixture was poured onto ice, washed with aqueous Na2CO3, extracted into benzene, and distilled in uacuo. Approximately 2 g of crude V was obtained as a viscous yellow liquid, bp 185" (0.1 mm). Mass spectroscopy showed that this product contained approximately 20% dinitrophosphonate, molecular ion m/e 346.0932 (calcd for C13HloN20~P,346.0928), as well as the desired mononitro compound, molecular ion m/e 301.1065 (calcd for C I ~ H Z O N O 301.1078). ~P, Thin layer chromatography on silica gel G F revealed at least five components when eluted with a mixture of ethyl acetate-10% absolute ethanol. The major spot proved to contain (mass spectroscopy) both the monoand dinitro materials. The entire sample was subjected to preparative tlc with recovery of the major spot. The partially purified material was chromatographed (also on silica gel GF) using the continuous elution techniqueI2 employing ethyl acetate-10% benzene. This approach afforded resolution of the two components, the dinitro compound exhibiting the larger Rt value. V was isolated as required by leeching from tlc plates. And. Calcd for C13HzoOsPN: C, 51.82; H, 6.69; N, 4.64; P, 10.28. Found:7 C, 51.82; H, 6.42; N, 4.85; P, 10.15. Electron Spin Resonance. Esr experiments a t Cornell University were carried out using the Varian V-4500 spectrometer system and the field-sweep calibration technique described previously.laJ* At Florida State University, esr measurements were made with a Varian E-12 spectrometer; the field sweep was calibrated with Fremi's salt in aqueous sodium carbonate solution. All the free radicals reported here were generated by electrolytic reduction ~ at 25" using the intra M U T O S ~technique. Assignments of coupling constants have been confirmed by comparison of the experimental spectra with spectra calculated using a slightly modified version of the computer program devised by Stone and Maki.16 This program was executed on an IBM 360-65 computer a t Cornell and on a CDC 6400 computer at Florida State University. Polarography. Polarographic measurements a t Cornell were made using the ORNL three-electrode polarograph described previously.13 A three-electrode polarograph constructed in our laboratory using all solid state Philbrick operational amplifiers and the Kelly-Fisher~Jonesl~current amplifier configuration was employed at Florida State University. All experiments were performed at approximately 25" and we report maximum, not average, diffusion currents. Potential measurements were made us. the aqueous saturated calomel electrode under conditions of "long immersion;"I8 the precision of these measurements is believed to be +5 mV. In all electrochemical procedures, tetraethylammonium perchlorate at 0.1 M concentration was employed as supporting electrolyte.
Results Polarography. Compounds I-V exhibited reproducible polarographic reduction waves which yielded the electrochemical parameters given in Table I. These results represent average values from at least three polarograms for each compound. Esr. Intra muros electroreduction of solutions ca. 1 mM in compounds I-V in both acetonitrile and dimethylformamide (DMF) produced strong esr signals which were attributable to the corresponding anion radicals. Esr data are summarized in Table I1 where we report average values of coupling constants obtained from the high- and low-field branches of two or more spectra. (11) G. M. Kosolapoff, ibid., 71, 1876 (1949). (12) E. Stahl, "Thin Layer Chromatography," M. R. F. Ashworth, translator, Springer-Verlag, New York, N. Y.,1969, p 69. (13) K. Kuwata and D. H . Geske, J . Amer. Chem. Soc., 86, 2101 ( 1964). (14) W. M. Gulick, Jr., and D. H. Geske, ibid., 88,4119 (1966). (IS) A.H.Maki and D. H. Geske, J. Chem. Phys., 30, 1356 (1961). (16) E. W. Stone and A. H. Maki, ibid., 38, 1999 (1963). (17) M.T. Kelley, H. C. Jones, and D. J. Fisher, Anal. Chem., 31, 1475 (1959). (18) D. H . Geske, J. L. Ragle, M. Bambenek, and A. L. Balch, J . Amer. Chem. SOC.,86,987 (1964).
1 January 12, 1972
31 Table I. Polarographic Results in Acetonitrile
I I1
1.153 1 180 1.086 I
111 IV V Nitrobenzene”
1.009 1.280 1.147
56 54 51 60 57 56
4.39 3.84 4.06 -2. 5a 4.35
Measurement complicated by apparent adsorption effect which leads to unusually short drop times ( ~ 1 . 5sec as compared with -6 sec for the other compounds). Data from ref 18. 0
1
6 Gauss
I
Table 11. Electron Spin Resonance Datao Anion radical
I
Acetonitrile aN
aH(m) ap
I1
aN aH(o)
ama
ap 111
aN
aH(o) ap
IV
aN
aH(o) aH(m) ap
V
aN
aH(o) UGHs
acH2 ap
Nitrobenzeneb
aN
aH(d aH(m) ~H(P)
DMF
10.40 3.43 1.12 6.07
f 0.06 f 0.07 f 0.02 f 0.06
9.88 3.46 1.12 6.35
f 0.07 f 0.05 f 0.01 f 0.08
10.58 3.36 1.04 9.32
f 0.05 f 0.01 f 0.01 f 0.06
10.00 3.37 1.04 9.66
f 0.04 f 0.02 f 0.01 f 0.04
10.00 f 0.06 3.35 f 0.01 10.00 f 0.06
9.53 f 0.03 3.33 f 0.02 10.49 f 0.04
8.48 3.28 0.96 24.80
f 0.05 f 0.01 f 0.02 f 0.11
7.78 3.23 0.90 27.15
f 0.03 f 0.01 f 0.01 f 0.11
11.06 3.33 1.04 2.29 19.14
f 0.05 f 0.03 f 0.02 f 0.02 f 0.07
10.29 3.36 1.01 2.34 20.09
f 0.06 f 0.02 f 0.02 f 0.01 f 0.07
10.32 3.39 1.09 3.97
9.83 3.32 1.09 3.97
Figure 1. Upper trace: low-field half of the esr spectrum of radical I obtained by electroreduction of an acetonitrile solution 0.7 mmol in compound I. Lower trace: computed spectrum using aN = 10.27, UH = 3.35, aH = 1.12, UP = 6.02, and line width = 0.125 G with Lorentzian line shape.
”‘It
I IO Gauss
I
Absolute values of coupling constants in gauss with uncertainties given at the 95 confidence level. Proton positions are indicated with respect to the nitro group. While g values were not determined exactly, all the radicals had g 2.0. *Acetonitrile data from D. H. Geske and A. H. Maki, J. Amer. C/zem. Soc., 82, 2671 (1960); D M F data from P. Ludwig, T. Layloff, and R. N. Adams, ibid., 86, 4568 (1964).
-
Figures 1-3 compare the experimental spectra of radicals I, 11, and V in acetonitrile solution with computed spectra. Comparable agreement was obtained for spectra in D M F solution. The computed spectra were obtained using the coupling constants measured from the individual spectra presented in the figures and those values are given in the figure captions. This procedure was necessitated by small variations in coupling constants among replicate spectra, a phenomenon reflected in the uncertainties quoted in Table 11. Assignment of spectra for radicals I11 and I V was straightforward and these spectra are omitted for brevity. For radical I11 in D M F all 18 lines are observable, whereas in acetonitrile this number is reduced to 12 by the accidental equivalence of uN and up. Computed spectra indicate that the equivalence of these coupling constants is at a level well within the uncertainty of the measurements. Although no interactions
Figure 2. Upper trace: low-field half of the esr spectrum of radical I1 obtained by electroreducton of an acetonitrile solution 0.8 mmol in compound 11. Lower trace: computed spectrum using uN = 10.55, aH = 3.365, aCHa= 1.04, ap = 9.24, and line width = 0.120 G with Lorentzian line shape.
with the protons of the m-phenyl groups were resolvable, such interactions probably account for the large line widths (0.34 G) observed in spectra of 111. The very large phosphorus coupling constant for radical I V permitted unique resolution of all 54 hyperfine lines when the solvent was DMF. In acetonitrile solution a small number of overlaps occurred, but assignment of the spectrum was still unambiguous. Gulick / Phosphorus Hyperfine Coupling Constants
32
1.1
t
IO Gauss
Figure 3. Upper trace: low-field section of the esr spectrum of radical V obtained by electroreduction of an acetonitrile solution 1.5 mmol in compound V. Slightly more than half the spectrum is shown; the arrow indicates the center. Lower trace: computed spectrum using the coupling constants given in Table 11, line width = 0.130 G , and Lorentzian line shape. Due to the large number of overlapped lines, these computations were very sensitive to small (10-50mG) changes in the two largest coupling constants.
Discussion Our previous investigation2 led us to suggest tentatively that electron spin density is transmitted from the nitrophenyl ring to the substituent phosphorus atom via a hyperconjugation-like mechanism. Direct interaction of orbitals on phosphorus with the pz orbital of the C 4 carbon atom in the aromatic ring would be required for such a mechanism. The spin density on phosphorus, and therefore the magnitude of the 31P hyperfine interaction, would be expected to depend upon the physical proximity of the phosphorus atom to this orbital. It is well establishedIg that for proton coupling constants in fragments of the form XCHzCH2the 0-proton coupling constant is expressible by eq 1, Here B and Bo
+
aH = ( B ~ ~ ( c o s 2e)),,.
(1)
are constants with B >> Bo and 8 is the angle between the C-H bond and the axis of the pz orbital associated with the paramagnetic center both projected on a plane perpendicular to the carbon-carbon bond. p” is the spin density at the Q position. In analogy to this treatment, we suggested the possibility of a relation such as eq 2 to describe 31Phypera p = C(cos2 O)pCaT
(2) fine splittings in a series of similar nitroaromatic phosphorus radicals.2 Here 8 is the angle between the axis of the pz orbital on C4 of the nitroaromatic ring and the P-X bond, both projected on a plane perpendicular to the X-C4 bond. This suggestion was based primarily on the observations that in acetonitrile up was 7.92 (19) R. Bersohn, J . Chem. Phys., 24, 1066 (1956); E. W. Stone and A. H. Maki, ibid., 37, 1326 (1962); 38, 1254 (1963).
G in diphenyl 4-nitrophenylphosphate (4 in ref 2; R = H, R’ = CeHsO, X = 0 in the present nomenclature) and 10.50 G in diphenyl 2,6-dimethyl-4-nitrophenyl phosphate ( 5 in ref 2; R = CH3, R’ = CeHsO,X = 0) and further, that in diethyl 4-nitrobenzylphosphonate (3 in ref 2; R = H, R’ = CzH60, X = CH2) the proton splitting for the bridging methylene group (2.70 G) was considerably smaller than that found in 4-nitrotoluene, 3.98 G. We felt that the first pair of observations implied a decrease in the angle 8, and the second indicated that since the phosphorus atom was sterically held away from the plane of the nitrophenyl ring (small 8) the steric environment of the methylene protons was constrained to be unfavorable for hyperconjugation. The mechanistic interpretation was of necessity tentative; in the first case we did not know with certainty the inductive effect at of 3J-dimethylation of the aromatic ring, while in the second we did not know the nature of the inductive effect to be expected from replacing one proton in 4-nitrotoluene with (C2H50)2P(0)-. The implication of eq 2 is that if, in a series of otherwise similar radicals, one goes from a situation in which there is free rotation of phosphorus about the X-C bond to a situation in which steric hindrance constrains 0 O”, the magnitude of up should double (the value of cos2 0 averaged over all values of 0 is 0.5). As can be seen in Table 11, this is approximately the case for the series 1-111; up increases from ca. 6 G in the unhindered radical I to cu. 10 G in 111. The data for this series are more complete than any previous data since we were able to prepare radicals in which both electron-releasing and electron-withdrawing groups were responsible for the steric perturbation. In accord with the stereochemical hypothesis, up continues t o increase from I1 to I11 as the size of the blocking group increases, despite their different inductive properties. Comparison of the nitrogen and ring proton coupling constants as well as the reduction potential for I with values for nitrobenzene suggests strongly that the (C6H&P(0)O- substituent has a negligible effect upon the electronic structure of the nitroaromatic ring. Differences observed between these same parameters in I1 and I11 and in nitrobenzene are such that they can be rationalized on the basis of effects anticipated18120 for the meta hydrocarbon substituents. Thus, since the necessary parameters for the phosphorus substituent are not well established, we feel that an estimate of the spin density at C4can reasonably be obtained from molecular orbital calculations for the corresponding metasubstituted nitrobenzene anions. Results of Hiickel calculations employing the McLachlan2 approximate configuration interaction method are given in Table 111. Using these values of pCIT we can proceed to estimate, albeit relatively crudely, values of C in eq 2 . The fact that the increase in up noted from I to I11 is less than twofold may be explained in several ways which may contribute in unknown proportions. The phosphorus group may not be entirely free to rotate in I, and/or residual torsional motions in I11 may prevent (cos2 0) from reaching its limiting value of unity. In addition, there may exist a spin polarization term analogous to Bo in eq 1 which is independent of 8.
-
(20) A. H. Maki and D. H. Geske, J. Amer. Chem. Soc., 83, 1852 (1961). (21) A. D. McLachlan, Mol. Phys., 3, 233 (1960).
Journal of the American Chemical Society / 94:l / January 12, 1972
33 Table 111. Molecular Orbital Calculations Radical anion" Nitrobenzene 3,5-Dimethylnitrobenzene* 3,5-Diphenylnitrobenzene. 3,4,5-Trimethylnitrobenzeneh
Spin densities Ci C2
0
N
7
Ct
Ca
-0.049 -0.047
0.168 0.169
0.245 0.196 0.002 0.143 -0.047
0.175
0.246 0,196 0.001 0.142
0.163
0.247 0.198 0.001 0.142 0.247 0.198 0.001 0.144
-0.046
Calculations performed on hydrocarbon substituted nitrobenzene anion radicals to estimate spin densities in the absence of established MO parameters for the phosphorus substituents; data here indicate little electronic perturbation on the nitroaromatic ring from these substituents (see text). MO parameters are those for acetonitrile solution from P. H. Rieger and G. K. Fraenkel, J . Chem. Phys., 39, 609 (1963). *Methyl groups included using parameters suggested by C. A. Coulson and V. A. Crawford, J. Chem. Soc., 2052 (1953). On the basis of molecular models, we estimate that the m-phenyl groups make an angle of about 60" with the nitrophenyl group and have employed a resonance integral appropriate for this angle. (1
Thus, we might wish t o formulate the phosphorus coupling constant as in eq 3. An upper limit to the
up
=
+ C(COS*
(co
mpc4c
(3)
contribution of Comay be determined by assuming that (cos2 0) has the values 0.5 and 1.0 respectively in I and 111. Using the acetonitrile data for I and I11 and the calculated spin densities with these assumptions, one obtains Co = 15.5 G and C = 41.6 G. Examination of space-filling molecular models, however, suggests that in I there is considerable steric hindrance of conformations in which the phosphorus atom lies in the plane of the nitroaromatic moiety. Hence, of the two assumptions above, the more realistic one is (cos2 0) = 1 in radical 111. If we ascribe the entire discrepancy between our data and the expected twofold increase in up t o steric hindrance in I (Le., Co = 0), then a value of C = 56 G is obtained. It seems certain that Co is less than 15.5 G and very likely that a hyperconjugative term with C 50 G dominates the behavior of u p in the series of radicals 1-111. Radical V provides the most convincing evidence supporting our assertion of the importance of steric effects. In the unmethylated analog ( 3 , ref 2; vide supra) we found up = 13.59 f 0.23 G in acetonitrile and 17.51 f 0.24 G in DMF, while the corresponding values of U C H ~were 2.70 f 0.02 and 2.81 f 0.02 G. Comparison of these data with results for radical V (Table 11) shows that in both solvents introduction of the m-methyl groups leads to a substantial increase in ap and a decrease in U C H ? well outside the limits of uncertainty in the coupling constants. Barring the unlikely circumstance that spin transmission to phosphorus involves some through space interaction with another part of the nitroaromatic ring, both up and aCH2 depend upon the same spin density, pChc. Thus, the changes of these coupling constants in opposite directions appear to prove the assertion of a steric effect; as conformations in which the phosphorus atom occupies positions for which 0 0" become preferred, the methylene protons are forced to occupy positions 60". We presume that radicals of this for which 0 type have a larger phosphorus splitting than those in
-
-
-
which the bridging atom, X, is oxygen, at least in part because of the more acute2 P-CH2-C angle (109' US. 120"). In analogy to the procedure followed for radicals 1-111, we obtain a rough estimate of pClc in V by noting that uN and Ell, are closely similar to values found in3,4-dimethylnitrobenzene anion18 and therefore simulate V with an MO calculation for 3,4,5-trimethylnitrobenzene anion (Table 111). On this basis (assuming Co = 0), a preliminary estimate of C for phosphonate radicals is 115 G. Radical IV is the first case in the 12 organophosphorus radicals studied here and in ref 2 in which we observe a substantial difference in spin density distribution from unsubstituted nitrobenzene anion. The most evident result of this electronic effect is a reduction of as by -2Oz from the value observed in nitrobenzene anion; ring proton splittings are also diminished slightly. The inductive effect of the substituent thiophosphate group is also reflected in the reduction potential for IV which is the most anodic of any of the organophosphorus free radicals we have studied in acetonitrile. Comparison of data for radical IV with those of the anion radical of 4-nitrobenzamide20 suggests -that the electronic effects of -C(O)NH2 and -SP(0)(OC8H& are exceedingly similar. Maki and ) Geske found aN = 8.37 G , U H ( " ) = 3.20 G, U H ( ~ = 0.98 G , and El,, = - 1.Ol4 V in the former case. It is interesting to note that both the electronic effects discussed in the preceding paragraph and the very large 31Psplitting observed for IV are specific effects of substitution of sulfur for oxygen ut the bridging position. Radical IV is closely related to the parathion anion radical (0,O-diethyl 0-4-nitrophenyl thiophosphate anion) reported in ref 2 for which the coupling constants in acetonitrile were uN = 10.27, u ~ ( =~ 3.47, ) a H ( m ) = 1.13, up = 10.29. Other previous data2 showed that up was substantially the same in radicals containing ethyl and phenyl ester groups, so we attribute the increase in up primarily to the movement of the sulfur atom from the phosphoryl position to the bridge position. The marked increase in the 31Pinteraction could be the result of a combination of increased conjugation through sulfur (the possibility of d-orbital participation exists) and also a geometrical factor since the P-S-C angle is likely to be more acutez2than the value of 120" observedz3for P-0-C. An X-ray crystallographic study of IV is in progress to investigate this latter point. Acknowledgments. Acknowledgment is made to the donors of the Petroleum Research Fund, administered by the American Chemical Society, for partial support of this research. Appreciation is also expressed for summer support during the initial phase of this work made possible by a grant from the Du Pont Co. to Cornell University. We thank the National Science Foundation for partial support for the purchase of the esr spectrometers through Grant GP-1687 to Cornell University and through Grant GP-17789 to Florida State University and for support of mass spectrometry (22) L. Pauling, "The Nature of the Chemical Bond," Cornell Universitypress, Ithaca, N. Y . , 1960, p 110; F. A . Cotton and G. Wilkinson, "Advanced Inorganic Chemistry," Interscience, New York, N. Y., 1966, pp 393, 394. (23) W. 0. Davies and E. Stanley, Acta Crystallogr., 15, 1092 (1962).
Gulick
Phosphorus Hyperfine Coupling Constants
34
facilities through Grant GU-2612 t o Florida State University. N I H support of the mass spectrometry facilities at Cornel1 University is acknowledged. We
thank Miss Vicki J. King for valuable technical assistance and Professor Paul Haake and Dr. Gail Hurst for a gift of compound I.
Solvent and Temperature Dependence of Hydrogen-Hydrogen, Hydrogen-Fluorine, and Fluorine-Fluorine Coupling Constants in Difluoroethylenes' Arthur M. Ihrig and Stanford L. Smith* Contribution from the Department of Chemistry, University of Kentucky, Lexington, Kentucky 40506. Received December 14, 1970 Abstract: Solvent- and temperature-induced variations of H-H, H-F, and F-F coupling constants in cis- and
trans-1,Zdifluoroethyleneand 1,l-difluoroethyleneare reported. These data, in conjunction with previous results, confirm that orientation of the solute dipole affects the magnitude of the solvent effect on geminal H-H and H-F coupling constants. Dipole orientation is not a factor in determining solvent effects on vicinal H-F and F-F The. coupling constants. The changes observed for vicinal H-F coupling constants are proportional to ~ K B dipole-induced solvent electric field is the primary causative factor producing the observed changes, but dispersion forces and specific interactions are also important. The best fit for a regression analysis of coupling constants against solvent parameters is of the form J o b s d = JO AR B D , where R and D represent reaction field and dispersion field terms. A and B are opposite in sign. Rough values of A H and A S are reported for weak solutesolvent complexes characterizing specific interactions effects. An apparent dichotomy is reported where one coupling constant in a molecule is solvent dependent but another is not, even though the same nuclei are involved in both couplings.
+
S
olvent-induced variations of various coupling constants ranging from 0.30 Hz to 10.0 Hz have been reported in several hundred compounds. The properties of the solute molecule and the solute-solvent interactions conducive to solvent-dependent coupling constants are the subject of much speculation. For the limited case of geminal H-H coupling constants, previous investigations3!5 have implicated the solute dipole as a necessary condition for solvent dependency. The observation that 'JHF of vinyl fluoride increases while 'JHF of trifluoroethylene decreases under similar conditions confirmed that the solvent effect depends on the presence of the solute dipole, and on its orientat i o n 4 In these cases it is assumed that the dipoleinduced solvent electric field (reaction field) alters the electronic distribution of the solute molecule, resulting in the observed changes in geminal coupling constants. However, reaction field effects are not the only mode of solute-solvent interaction. Laszlo6 has demonstrated an excellent correlation between dispersion forces and coupling constant changes in symmetric molecules. Recently, Goldstein7 has developed and tested a model involving both reaction field terms and equilibrium constants for collision complexes in order to explain (1) This work supported in part by DOD Project Themis Contract No. DAAB07-69-C-0366, A portion of this work was presented at the 154th National Meeting of the American Chemical Society, Chicago, Ill., Sept 1967. (2) References 3 and 4 contain many citations for early reports of solvent-dependent coupling constants, (3) S. L. Smith and R. H. Cox, J . Chem. Phys., 45, 2848 (1966). (4)S. L. Smith and A. M . Ihrig, ibid., 46, 1181 (1967). ( 5 ) V . S. Watts and J . H. Goldstein, ibid., 42, 228 (1965). (6) P. Laszio and A. Speert, J . Mugn. Resonance, 1, 291 (1969). (7) R . L. Schmidt, R. S. Buttler, and J. H. Goldstein, J . Phys. Chem., 73, 1117 (1969).
Journal ojthe American Chemical Society
94:l
+
discrepancies in extrapolating reaction field terms to high dielectric constant solvents. The difluoroethylenes provide a number of unique opportunities to clarify some of these problems. The cis isomer possesses a permanent dipole whose orientation relative to the H-C-F group is intermediate to that of vinyl fluoride and trifluoroethylene, while the trans isomer contains the same H-C-F group but has no dipole moment. The dipole moment-solvent electric field concept predicts that AJ,,,S for cis-l,2-
difluoroethylene should be between the values observed for vinyl fluoride and trifluoroethylene, while 'JHF of trans-1 ,2-difluoroethylene should be solvent invariant. All three isomers have a nice array of vicinal H-F couplings. Previous indicate that solvent effects on vicinal H-F coupling constants are at least an order of magnitude larger than solvent effects on the corresponding vicinal H-H couplings (which are generally insensitive to solvent). The 1,2-difluoroethylenes present an excellent opportunity to compare the solvent-induced changes of vicinal H-F coupling constants with the corresponding changes in geminal (8) A positive AJ is defined as a change for which J increases algebraically with enhanced solvent polarity. (9) H. M. Hutton and T.Schaefer, Can. J . Chem., 45, 1111 (1967).
January 12, 1972
